RESUMO
Although still in its infancy, artificial intelligence (AI) analysis of kidney biopsy images is anticipated to become an integral aspect of renal histopathology. As these systems are developed, the focus will understandably be on developing ever more accurate models, but successful translation to the clinic will also depend upon other characteristics of the system.In the extreme, deployment of highly performant but "black box" AI is fraught with risk, and high-profile errors could damage future trust in the technology. Furthermore, a major factor determining whether new systems are adopted in clinical settings is whether they are "trusted" by clinicians. Key to unlocking trust will be designing platforms optimized for intuitive human-AI interactions and ensuring that, where judgment is required to resolve ambiguous areas of assessment, the workings of the AI image classifier are understandable to the human observer. Therefore, determining the optimal design for AI systems depends on factors beyond performance, with considerations of goals, interpretability, and safety constraining many design and engineering choices.In this article, we explore challenges that arise in the application of AI to renal histopathology, and consider areas where choices around model architecture, training strategy, and workflow design may be influenced by factors beyond the final performance metrics of the system.
Assuntos
Inteligência Artificial , Confiança , Humanos , RimRESUMO
PURPOSE: Modern strategies for the screening and diagnosis of Gestational Diabetes Mellitus (GDM) rely on universal Oral Glucose Tolerance Test (OGTT). However, they are unsustainable in low-income countries. In this study, we aimed at assessing the feasibility of a simplified diagnostic policy. METHODS: The study took place in an urban referral hospital in Freetown, Sierra Leone. During an 11-month period, pregnant women were offered capillary blood test for glucose assessment. They could be screened at any time during pregnancy. GDM was diagnosed if fasting glucose was ≥ 92 mg/dl or if the OGTT was positive. The latter was prescribed only to women presenting after 24 weeks' gestation with at least one risk factor for GDM and fasting capillary glucose between 85 and 91 mg/dl. A definitive diagnosis required confirmation to this aim, women with values above the thresholds were invited to refer the next working day for repeating the test after fasting overnight. RESULTS: Overall, 7827 women were referred for screening, of whom 6872 (87%) underwent at least one capillary glucose assessment. However, 895 of those who had a positive test did not return for confirmation. Overall, a definite assessment could be done in 5799 subjects corresponding to 76% (95% CI 75-77%) of those eligible. GDM was diagnosed in 128 women (1.9%, 95% CI 1.6-2.2%). Based on an expected confirmation rate of 22% (calculated from those who referred for confirmation) in the 895 women who did not come back, one could infer that GDM would have been diagnosed in additional 197 women, raising the prevalence to 4.7% (95% CI 4.2-5.3%). CONCLUSION: Three quarters of subjects could be assessed with our approach. Data also suggest that GDM is not rare even if identification of affected cases remains challenging.
